pointcloud-analysis

Codes for the Point Cloud Curvature and Saliency Analysis Project.

• Image Processing Lab, Sharif University of Technology

Note: In progress, new implementations regarding to this project will be added soon.

Structure

• src/pointnet: Contains PyTorch implementation of the Pointnet model and all the related utility functions to train and evaluate it on the ModelNet data.
• src/dgcnn: Contains PyTorch implementation of the DGCNN model and all the related utility functions to train and evaluate it on the ModelNet data.
• src/saliency_map.py: Contains the PyTorch implementation of the Point Cloud Saliency Maps.
• src/deep_fool.py: Contains code for the DeepFool method, a simple and accurate method to fool deep neural networks (Pytorch & PointCloud Implementation).

Models

Model checkpoints are available here.

Datasets

ModelNet Dataset from Priceton ModelNet.

References:

• https://github.com/aryanmikaeili/Pointnet-CW-attack
• https://github.com/tianzheng4/PointCloud-Saliency-Maps

Citation

Please cite us if you use our work in your research.